middleware-ops

name	middleware-ops
description	Deploy and operate Redis, MySQL, Kafka, PostgreSQL and other middleware with Helm, kubectl, and vendor CLIs. Use for installation, upgrades, status checks, scaling, performance tuning, backup planning, database or cache incidents. 适用于中间件部署、Redis/MySQL/Kafka 运维、数据库状态、缓存问题和性能排查。

Middleware Deployment and Operations Guide

Redis Deployment

Helm Deployment

helm install redis bitnami/redis \
  --namespace middleware --create-namespace \
  -f charts/middlewares/redis/values.yaml

Key Config

Parameter	Description	Default
architecture	standalone/replication	replication
auth.password	Auth password	Random generated
master.persistence.size	Master storage	8Gi
replica.replicaCount	Replica count	3
master.resources.limits.memory	Memory limit	256Mi

Health Check PromQL

# Redis connections
redis_connected_clients

# Redis memory usage
redis_memory_used_bytes / redis_memory_max_bytes * 100

# Redis hit rate
rate(redis_keyspace_hits_total[5m]) / (rate(redis_keyspace_hits_total[5m]) + rate(redis_keyspace_misses_total[5m])) * 100

# Redis slow queries
redis_slowlog_length

MySQL Deployment

Helm Deployment

helm install mysql bitnami/mysql \
  --namespace middleware \
  -f charts/middlewares/mysql/values.yaml

Key Config

Parameter	Description	Default
architecture	standalone/replication	standalone
auth.rootPassword	root password	Random generated
primary.persistence.size	Primary storage	20Gi
primary.resources.limits.memory	Memory limit	512Mi

Health Check PromQL

# Connection usage
mysql_global_status_threads_connected / mysql_global_variables_max_connections * 100

# QPS
rate(mysql_global_status_queries[5m])

# Slow query count
rate(mysql_global_status_slow_queries[5m])

# Lock wait
mysql_global_status_innodb_row_lock_waits

Kafka Deployment

Helm Deployment

helm install kafka bitnami/kafka \
  --namespace middleware \
  -f charts/middlewares/kafka/values.yaml

Key Config

Parameter	Description	Default
replicaCount	Broker count	3
persistence.size	Storage size	50Gi
kraft.enabled	Use KRaft mode	true
resources.limits.memory	Memory limit	1Gi

Health Check PromQL

# Consumer Lag
kafka_consumergroup_lag_sum

# Active brokers
count(kafka_server_brokertopicmetrics_messagesin_total)

# Topic message rate
sum(rate(kafka_server_brokertopicmetrics_messagesin_total[5m])) by (topic)

Middleware General Operations

Post-Deployment Verification Checklist

1. All Pods Running and Ready
2. Service and Endpoint correct
3. PVC bound with correct capacity
4. Connection test passes (internal + external)
5. Monitoring metrics reporting
6. Backup policy configured (where applicable)

Scale Operations

• Check node resource headroom before scale-up
• Stateful service scale-up needs storage class support
• Confirm data migrated/backed up before scale-down
• All scale operations require user confirmation

Troubleshooting Flow

1. Check Pod status and events
2. Check resource usage (CPU/memory/disk)
3. Query recent logs for errors
4. Check if connections at limit
5. Check if storage full
6. Check network policies blocking

Safety

• Render and diff Helm changes before upgrade; pin chart versions.
• Require approval before install, upgrade, rollback, scale, failover, restore, or delete.
• Back up stateful data and document rollback before risky changes.
• Retrieve credentials from Kubernetes Secrets only when required and never print them.